Security Alert: 15 JetBrains plugins steal IA API keys and monetize them

An independent research team has discovered a coordinated campaign that infiltrated at least 15 malicious supplements into the JetBrains Marketplace with the specific objective of steal API keys from IA services introduced by developers in the configuration of those plugins. The Aikido Security technical report documents that the supplements - presented as encoding assistants, code reviewers and Git utilities - were apparently functioning legitimately, but contained code that sent the credentials to the remote server as soon as the user pressed "Apply" after sticking its key to the settings.

The campaign's mechanics is restless simple and effective: the keys were exfiltered by HTTP to an encoded endpoint (hxxp: / / 39.107.60 [.] 51 / api / software / key), which means that any organization that had such plugins installed could see how their credentials were filtered without encryption or control. Aikido found that the packages shared similar code blocks and that the activity began in October 2025, with new publications until June 2025, accumulating about 70,000 downloads reported among all affected plugins, with two of them exceeding 25,000-27,000 downloads according to the available counters.

In addition to exfiltration, researchers observed an additional and worrying functionality: a payment "tier" that, after paying a small amount, led the remote server to return an IA supplier key for the plugin to use on behalf of the user. This suggests a model where stolen keys to unanticipated users are redistributed or monetized, a behavior that has no legitimate justification as serious suppliers do not deliver active and unlimited keys to third parties in that way.

The security implications are direct and serious. A compromised API IA key allows an attacker to generate costs in the supplier's account, access to potentially confidential histories and prompts, evade organizational controls and, in some cases, use models for automated scanning, payload generation or sensory data abuse. The risk includes fraudulent billing, leakage of sensitive information and abuse of large-scale models in addition to the potential for the sale of these keys in clandestine markets.

For developers and equipment using IDes JetBrains, the recommended immediate actions are clear: if you installed any of the identified or suspicious supplements, uninstall it and break the affected keys immediately. Do not wait for additional tests: change the keys from the IA supplier panel, enable domain, IP or use limits if the supplier allows, and activate abnormal billing and consumption alerts. Then check the use records to detect calls from unusual locations, schedules or volumes.

In parallel, it applies operational measures to reduce exposure in the future: it avoids introducing API keys directly into third-party plugin settings; it prefers environment variables, computer-managed configuration files or delegated mechanisms such as OAuth and tokens with reduced permissions. For organizational accounts, use company-run credentials with permissions and separate billing limits, and consider use gateways or API proxies that can inspect and block suspicious traffic. For guidance on good API security practices and credentials management, resources such as the OWASP APIs security project offer practical recommendations: OWASP API Security Project.

It is also up to ecosystem maintainers to take action: marketplaces should improve the static and dynamic review of plugins, require signatures and identity verification processes of editors, and provide rapid channels of response and removal of malicious supplements. Meanwhile, security teams should monitor plugin facilities in development endpoints and block known exfiltration patterns at the network level. You can review and audit the additions available in the official JetBrains marketplace at plugins.jetbains.com and report suspicious behaviour to those responsible for the repository and to incident response groups.

If you suspect that you were affected, in addition to rotating keys and reviewing logs, please inform your IA provider and the internal security team; consider submitting a report to incident response entities or Aikido itself, which published a detailed analysis of the case and help understand the commitment indicators: Aikido Security - report. It is also good practice to block at the network level the identified endpoint if possible and to keep evidence for forensic support.

This incident highlights a broader lesson: as the integration of IA into development tools accelerates, confidence in third-party extensions must be accompanied by technical and operational controls. Auditing the software we use to program and apply the principle of lesser privilege in our service keys is no longer optional, it is an essential measure to protect both work and budgets and organizational privacy.